Next Generation IoT (NG-IoT) architectures and applications are cornerstone to radical innovations in various applications domains, such as industry 4.0, smart transportation, health or agriculture. Key enablers for NG-IoT are hyper-connectivity (5G+), edge-cloud computing and AI/ML, leading to the generalization of knowledge creation at the edge but also to growing autonomy of edge entities. Supporting NG-IoT requirements, 5G sidelink communication enables device-to-device and soon even multi-hop ad hoc communications. Edge-cloud platforms provides processing, storage and intelligence closer to IoT devices or even integrated with IoT devices. And recent federated or gossip learning mechanisms pave the way to cooperative knowledge creations between edge entities. With cooperative knowledge creation between edge-cloud platforms over peer-to-peer wireless links comes the challenge of knowledge segmentation, where knowledge location is not unique and where various versions of a given knowledge may be available in different locations or time. Knowledge at the edge therefore opens the era of knowledge networking. This keynote first highlights NG-IoT cooperative decentralized innovations, covering 5G sidelink and ad hoc multi-hop networking, edge-cloud and dew computing, and federated/gossip machine learning. It then introduces a framework for knowledge-centric networking, emphasizing challenges behind autonomously naming and creating knowledge or identifying and locating knowledge. It finally illustrates the impact of the proposed framework with selected use cases of knowledge networking applied to connected vehicles.
Knowledge on the edge - From information to knowledge networks
WONS 2021, Keynote speech, 16th Wireless On-demand Network Systems (Virtual Conference)
Systèmes de Communication
© 2021 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
PERMALINK : https://www.eurecom.fr/publication/6490